Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
VPC endpoints let resources reach cloud services over the provider’s internal network instead of the internet, and endpoint policies can restrict which accounts and buckets they may reach.

What Are VPC Endpoints and Private Link?
A VPC endpoint is a private entry point to a cloud service from inside your own virtual network. Without one, an instance calling object storage or a queue resolves a public hostname and reaches it through an internet gateway or NAT gateway, even though both ends belong to the same provider. With one, the traffic stays on the provider’s internal network, the resource needs no route to the internet, and you gain a policy point where you can state which services, accounts, and resources may be reached.
| Gateway endpoint | Interface endpoint | |
|---|---|---|
| Mechanism | A prefix list entry added to route tables | An elastic network interface with a private IP in your subnet |
| Services | S3 and DynamoDB only | Most AWS services, plus partner and private services |
| Cost | No charge | Hourly charge per endpoint per availability zone, plus data processing |
| Reachable from on-premises | No | Yes, over Direct Connect or VPN |
| Technology | Routing | AWS PrivateLink |
PrivateLink also works in the other direction. A software vendor can publish an endpoint service backed by a network load balancer, and customers create interface endpoints to consume it, so the two parties exchange traffic without peering networks, sharing route tables, or exposing anything publicly. For SaaS providers serving regulated customers this is often the difference between winning and losing the security review.
The other providers offer the same shape under different names. Azure Private Endpoint assigns a private IP in your virtual network for a specific resource, and Azure Private Link is the service behind it; the older service endpoints keep the traffic on the Microsoft backbone but do not give the resource a private address. Google Cloud offers Private Service Connect alongside Private Google Access.
Keeping traffic off the public internet is useful, but it is not the main benefit — service calls were already encrypted in transit. The real gain is that an endpoint is a chokepoint you can write rules at.
An endpoint policy restricts what may be done through that endpoint regardless of what the calling identity is otherwise permitted to do. A policy allowing access only to buckets inside your own organisation, expressed with a condition on the organisation identifier, means that a compromised instance cannot copy data to a bucket the attacker controls. Without it, the endpoint faithfully carries exfiltration to any bucket in the world, because object storage is a global namespace.
The reverse condition is equally valuable. A bucket policy that denies requests unless they arrive through a named endpoint, or from a named VPC, ensures that leaked credentials are useless outside your network. Used together, the two conditions give you something close to network-scoped data access:
{
"Effect": "Deny",
"Principal": "*",
"Action": "s3:*",
"Resource": ["arn:aws:s3:::finance-exports/*"],
"Condition": {
"StringNotEquals": {"aws:SourceVpce": "vpce-0a1b2c3d4e5f"}
}
}
Private endpoints belong with the other controls that remove implicit network trust: subnets with no internet route, security groups scoped to specific endpoints, and egress inspection for whatever genuinely must leave. In a zero trust design the network is not the authorisation mechanism, but it remains a useful constraint — and the combination of endpoint policy with identity policy produces defence in depth that neither provides alone.
For organisations handling payment data, keeping storage and database traffic off public networks simplifies PCI DSS scoping arguments considerably, and the NCSC Cloud Security Principles on separation and interface protection point in the same direction. If you want to know whether your private networking holds under pressure, our network security services test the routes that exist rather than the ones that were designed.
They do not add encryption. Calls to cloud service APIs already use TLS, so the data was encrypted in transit before you added an endpoint. What changes is the path the traffic takes, which no longer traverses the public internet, and the fact that you now have a place to apply an endpoint policy. Treat endpoints as a routing and policy control rather than an encryption one.
Only with a restrictive endpoint policy attached. By default an endpoint allows access to any resource in that service, including storage buckets belonging to an attacker, so it carries exfiltration as happily as legitimate traffic. Add a condition restricting access to resources within your own organisation, and pair it with resource policies that reject requests arriving from outside your endpoints.
Almost always DNS. Interface endpoints depend on a private hosted zone that resolves the service hostname to the endpoint’s private address. If private DNS is disabled on the endpoint, or the hosted zone is not associated with the virtual network making the request, the public name resolves normally and traffic takes the internet route. Check resolution from inside the subnet rather than from your workstation.
For workloads handling sensitive data, usually yes, because they let you remove internet routing entirely and apply endpoint policies. For small or non-sensitive environments the hourly charges across many services and availability zones can outweigh the benefit. A common compromise is to centralise endpoints in a shared services account, share them across the organisation, and reserve full coverage for regulated workloads.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.